TWI779183B - Systems, methods, and media for artificial intelligence feedback control in additive manufacturing - Google Patents

Abstract

Additive manufacturing systems using artificial intelligence can identify an anomaly in a printed layer of an object from a generated topographical image of the printed layer. The additive manufacturing systems can also use artificial intelligence to determine a correlation between the identified anomaly and one or more print parameters, and adaptively adjust one or more print parameters. The additive manufacturing systems can also use artificial intelligence to optimize one or more printing parameters to achieve desired mechanical, optical and/or electrical properties.

Description

Artificial intelligence feedback control system, method and medium in additive manufacturing

This article is about mechanisms used to provide artificial intelligence feedback control in additive (additive) manufacturing.

Additive manufacturing systems, such as 3D printers and cell printers, are widely used to deposit multiple layers of natural, synthetic or biological materials for fabrication through processes of extrusion, sintering, photopolymerization, mechanosynthesis or electrohydrodynamic forces object. The process of additive manufacturing creates objects through a layer deposition process, where an additive manufacturing printer keeps adding successive layers until the printed object is complete.

Typically, objects printed by additive manufacturing printers are based on production designs. Three-dimensional modeling software (eg, a CAD program) can be used to create a production design of desired specifications for the object. A layer-slicing program can then convert the production design into a numerical control code (e.g., G-code) that divides the design into layers, which can then be used to instruct an additive manufacturing printer to print each individual layer of the production design entity representation. The goal of additive manufacturing is to print an object that closely matches the specifications of the production design.

Depending on the size and complexity of the production design, printed objects can take hours to days to complete. Current additive manufacturing systems are limited in the type of feedback they can provide and the corrective actions they can take after printing each layer of an object. Usually no feedback is provided until the entire object has been printed. When providing feedback during the printing process of an object, it is typically used for the purpose of determining whether to stop or continue printing the object.

In some additive manufacturing systems, feedback is provided by shadows produced by printed objects when light falls on them. This approach is limited because shadows can obstruct many areas of the printed object and prevent accurate feedback. Accurate feedback in additive manufacturing is useful to maintain quality and reproducible printed objects.

Therefore, it is desirable to provide artificial intelligence feedback control (AIFC) for each printing layer of an object so that timely corrective actions can be taken during the printing process of the object. It is also desirable to provide AIFCs to achieve desired mechanical, optical and/or electrical properties of printed objects, and to achieve printed objects that closely resemble their production designs, or to improve production designs.

According to some embodiments, systems, methods, and media for artificial intelligence feedback control in additive manufacturing are provided. More particularly, in some embodiments, an additive manufacturing system is provided, the system comprising: a printing head configured to print an object in a layer-by-layer manner; for providing a surface of a printed layer of the object with an illumination source for illumination; an image sensor configured to capture an image of the printed layer; and at least one hardware processor configured to: receive the captured image; generate a three-dimensional topography of the printed layer image; using a first artificial intelligence algorithm, identifying an anomaly in the print layer from the generated topography image; using a second artificial intelligence algorithm, determining a correlation between the identified anomaly and one or more printing parameters ; and assign values to one or more print parameters.

In some embodiments, a method for additive manufacturing is provided, the method comprising: receiving a captured image generated by an image sensor configured to capture to sequentially The image of the printing layer of the object printed in layer mode; using a hardware processor to generate a three-dimensional topography image of the printing layer; using the first artificial intelligence algorithm to identify the printing layer from the generated topography anomalies; using a second artificial intelligence algorithm, determining a correlation between the identified anomalies and the one or more printing parameters; and assigning values to the one or more printing parameters.

In some embodiments, there is provided a non-transitory computer-readable medium containing a plurality of computer-executable instructions that, when executed by a processor, cause the processor to execute a method for additive manufacturing A method comprising: receiving a captured image generated by an image sensor configured to capture an image of a printed layer of an object printed in a layer-by-layer manner ; generating a three-dimensional topography image of the printing layer; using a first artificial intelligence algorithm to identify anomalies in the printing layer from the generated topography image; using a second artificial intelligence algorithm to determine whether the identified anomalies are consistent with a or associations between printing parameters; and assigning values to one or more printing parameters.

According to some implementations of the disclosed subject matter, various mechanisms (which may include systems, methods, apparatuses, devices, etc.) for additive manufacturing artificial intelligence feedback control (AIFC) are provided. For example, AIFC may be useful for optimizing printing parameters of an additive manufacturing system to achieve desired mechanical, optical and/or electrical properties and/or a desired accuracy of printed objects as compared to production design. AIFC can also be used to identify anomalies in the print layer and take corrective action during the printing process.

As disclosed herein, in some embodiments, artificial intelligence can be used to learn and improve from additive manufacturing as described herein and output feedback, information, data, and/or instructions (“AIFC”). Artificial intelligence algorithms may include one or more of the following alone or in combination: machine learning, hidden Markov models, recurrent neural networks, convolutional neural networks neural networks), Bayesian symbolic methods, general adversarial networks, support vector machines, and/or any other suitable artificial intelligence algorithm. Although AIFC is based on AI algorithms, AIFC can also use data collected during additive manufacturing that is not based on AI algorithms.

FIG. 1 shows an exemplary additive manufacturing system 100 that can implement AIFC, according to some implementations of the disclosed subject matter. At a high level, the basic elements of the additive manufacturing system 100 include a digital control code generator 110, an additive manufacturing printer 115, an image generator 170, and an image analyzer 180, according to some embodiments. The additive manufacturing printer 115 may include an image sensor 120 , a light source 130 , a printing head 140 , a wire supply system 145 , a bottom plate 150 and a control module 160 . The functions of the elements used in the additive manufacturing system 100 may be combined in a single element or distributed among several elements. In some embodiments, the functions of some components (eg, digital control code generator 110 , image generator 170 , and/or image analyzer 180 ) may be performed remotely from additive manufacturing printer 115 .

Note that additive manufacturing system 100 may include other suitable elements not shown. Additionally or alternatively, some elements included in the additive manufacturing system 100 may be omitted.

Although the following description refers to the use of AIFCs with fused deposition modeling additive manufacturing printers, in some embodiments, the AIFCs described herein may be used with any suitable 3D printing technology, including stereolithography (SLA), Electron beam melting, direct metal deposition (electrohydrodynamic printing) and selective laser sintering.

In some embodiments, the additive manufacturing printer 115 may include one or more image sensors 120 for capturing images and/or video during the printing process. The image sensor 120 may be configured to capture an image (or video) of the object while and/or after printing each layer of the object. The image sensor 120 may be, for example, a charge-coupled device (CCD) or a complementary metal-oxide-semiconductor (CMOS) sensor, such as may be used in a digital camera and/or a video camera. The image sensor 120 may also include an infrared (IR) camera for capturing thermal images and/or video of objects and performing temperature calculations. The image sensor 120 can be located at different positions and angles relative to the bottom plate 150 and/or the print head 140 .

In some embodiments, additive manufacturing printer 115 may include a single light source 130 or multiple light sources (e.g., multiple light vectors) positioned at different positions and angles relative to chassis 150 and/or relative to image sensor 120 (eg, light sources may be placed around the periphery of the image sensor 120). Illumination may vary according to size, number of light sources used and/or position and angle of illumination. Illumination can be used to illuminate the printed layer of the object so that the image sensor 120 can capture an image and/or video of the object.

Captured images and/or videos may be stored in memory and may be used to create a three-dimensional topographical image of the printed layer and/or other suitable images, as discussed herein in connection with image generator 170 .

FIG. 2 shows an example of a camera 120 and light source 130 that may be used with an additive manufacturing printer 115 . FIG. 2 includes the camera 120 surrounded by a light emitting diode (LED) ring 210 placed in an LED holder 230 . In some embodiments, the control module 160 controls the individual LEDs within the LED ring 210 to determine which LEDs should be illuminated. The control of the individual LEDs within LED ring 210 may be determined by the requirements of the topographical imaging technique used to generate the printed layer image.

As noted above, the additive manufacturing printer 115 may also include one or more print heads 140 and one or more chassis 150 . Print head 140 and/or chassis 150 are movable relative to each other in X (width), Y (length) and Z (height) dimensions. The print head 140 can hold the filament supplied by the filament supply system 145 , and the filament is extruded layer by layer through one or more nozzles of the one or more print heads 140 . In some embodiments, the temperature of the printhead nozzles can be controlled to heat the filament stored in the printhead 140 so that the filament remains in a flowable form where it can be deposited (e.g., when the printhead 140 and/or the chassis 150 relative to one another, and/or when print head 140 and/or base plate 150 are stationary). The extruded material may be fused to the base plate 150 (as in the case of the first extruded layer) or to a previously deposited extruded layer. Other aspects of print head 140 and/or chassis 150 that may be controlled include, for example, the path that print head 140 and/or chassis 150 follows during movement, the print head and/or The amount of movement of the base plate 150 relative to the other along the Z-axis dimension, the orientation of the print head 140 and/or the base plate 150 relative to the other, the speed of movement of the print head 140 and/or the base plate 150, and the deposition of the filament by the print head 140 amount and rate. In some embodiments, the printing path can be defined by at least two sets of X-Y-Z coordinates. During operation, the print head and/or the chassis can be controlled to move relative to the other, and the print head can release the filaments in a desired fill pattern. In some embodiments, the print head 140 and/or the chassis 150 may be controlled by codes from the digital control code generator 110 and/or the control module 160 .

In some embodiments, the bottom plate 150 can be heated to a predetermined temperature and can be oriented in different directions.

In some embodiments, the control module 160 (which may include a controller and a controller interface in some embodiments) can control the components of the additive manufacturing system 100 (for example, the digital control code generator 110, the image sensor 120, any suitable one or more settings (e.g. temperature, speed, direction, etc.) printing an image of the object, enabling the light source 130, etc.), and calculations performed by and between components of the AM system. Control system 108 may include any suitable hardware (and in some embodiments may execute software), such as computers, microprocessors, microcontrollers, application specific integrated circuits (ASICs), field programmable gate arrays (FGPAs), and Digital signal processors (DSP) (any of which may be referred to as a hardware processor), encoders, circuits for reading encoders, memory devices (including one or more EPROMS, one or more EEPROMs, dynamic random memory memory (“DRAM”), static random access memory (“SRAM”) and/or flash memory), and/or any other suitable hardware components. In some embodiments, each component within the additive manufacturing system 100 may include its own software, firmware, and/or hardware to control each component and communicate with other components in the additive manufacturing system 100 .

In some embodiments, communication between the control module 160 and other components of the additive manufacturing system 100, and/or communication between the control module 160 and other components within the additive manufacturing printer 115 may use any suitable Communication technologies such as analog technologies (e.g., relay logic), digital technologies (e.g., RS232, Ethernet or wireless), network technologies (e.g., local area network (LAN), wide area network (WAN), Internet road), Bluetooth technology, near field communication technology, secure RF technology and/or any other suitable communication technology.

In some embodiments, operator input may be communicated to control module 160 using any suitable input device (eg, keyboard, mouse, or joystick).

In addition to the elements shown in Figure 1, the additive manufacturing printer 115 may include other elements such as temperature sensors, humidity sensors, sensors for measuring acceleration and any unintended movement of the print head 140 ( such as shaking, trembling, etc.), and a display monitor for displaying images. Additive manufacturing printer 115 may also include one or more actuators for orienting and/or moving image sensor 120 , illumination source 130 , print head 140 and/or chassis 150 .

Figure 1 also shows a digital control code generator 110 . In some embodiments, digital control code generator 110 may be configured to receive a three-dimensional design (eg, a computer-aided design (CAD) model) of an object to be printed (referred to herein as a "production design"). The production design may be received in any suitable format capable of being processed by the digital control code generator 110 (eg, standard tessellation language (.stl), drafting standard (DWS), or drafting (DWG) file format).

Digital control code generator 110 may be configured to convert the production design into instructions for additive manufacturing printer 115 to print a physical representation of the production design. In some embodiments, as described in conjunction with FIG. 3 , the digital control code generator 110 may include an interface for the operator to input certain printing parameters. The printing parameters may also include, but are not limited to, one or more of the following: printing characteristics of the additive manufacturing machine 115 (eg, print head size, extruded filament type, 3D printing technology, etc.); printing path; filament Supply rates; specifications for production designs (eg, what the design should look like when printed, desired mechanical, optical and/or electrical properties, etc.).

Based on one or more printing parameters, the digital control code generator 110 may apply a slice algorithm to intersect the production design with parallel planes spaced a predetermined distance apart in the Z direction to generate two-dimensional or three-dimensional layers. For example, if an object to be printed is 5mm in the Z direction and the desired layer thickness is 0.2mm in the Z direction, then the production design of the object can be sliced into 25 layers that are 0.2mm thick in the Z direction. In addition to cutting the production design, the digital control code generator 110 can also be configured to generate a digital control code for each layer to be printed based on one or more of: printing parameters; AIFC of one or more previously printed layers; AIFC from other printed objects (some of which may be incorporated into the production design); and the printing characteristics of the additive manufacturing printer 115.

In other embodiments, the layer-slicing algorithm may be configured to only determine the first layer and generate the digital control code for that first layer. The numerical control codes for each subsequent layer of the printed object may be generated based on one or more of: AIFC from one or more previously printed layers of the currently printed printed object; AIFC (some of which may be incorporated into the production design); desired mechanical, optical and/or electrical properties and specifications of the production design for the printed object; input parameters entered by the operator and/or additive manufacturing printer 115 printing characteristics. In some embodiments, the slice algorithm may be omitted entirely, and the numerical control code may be generated based on one or more of: AIFC from other printed objects (some of which may be incorporated into the production design); Desired mechanical, optical, and/or electrical properties and specifications of the production design of the object; input parameters entered by the operator; and/or printing characteristics of the additive manufacturing printer 115 . In some embodiments, the digital control code generator may also take into account uncontrollable variables (i.e., uncontrollable variables that do not require human intervention), including, but not limited to, environmental humidity, temperature, and lighting of the additive manufacturing printer 115, Voltage variations, wear, and bus supply available to printhead 140 .

FIG. 3 shows an example interface 300 for an operator to enter printing parameters into the digital control code generator 110 according to some implementations of the disclosed subject matter.

Note that interface 300 may include fields for controlling other suitable printing parameters not shown. Additionally or alternatively, some printing parameters included in interface 300 may be omitted in some embodiments. Furthermore, any and all printing parameters included in interface 300 may alternatively be automatically generated by a digital control code generator rather than input by an operator. In some embodiments, an operator may input printing parameters for producing the first layer of the design, and the digital control code generator 110 may use artificial intelligence algorithms and other methods disclosed herein to generate subsequent layers for printing the design. The printing parameters for the layer.

As shown, interface 300 may include fields to control print quality parameters such as layer height, shell thickness, and retraction.

The layer height refers to the height of the layer of the printed object. Layer height affects print resolution. For example, short layers can produce more detailed prints and smoother surfaces than higher layers. However, objects with shorter layers may take longer to print. Conversely, higher layers may correspond to lower resolution printing and rougher surfaces. Objects that include high layers can be printed faster than the same object with shorter layers. Any suitable layer heights may be used in some embodiments.

Shell thickness refers to the thickness of the outer wall of the printed object. Any suitable shell thickness may be used in some embodiments.

Retraction refers to ensuring that no filament is extruded from the print head when the print head moves to an area that is not designated for printing. In some embodiments, retraction can be enabled or disabled.

Interface 300 may also include fields to control print speed settings for controlling print head speed and/or platen speed. Printing speed refers to the speed at which the print head and/or the platen move while the print head is printing. Any suitable printing speed may be used in some embodiments. Based on the printing speed, the amount of material that needs to be extruded (ie, the feed rate) can be calculated. Any suitable feed rate may be used in some embodiments.

Interface 300 may also include fields to control temperature settings for controlling print head temperature and/or chassis temperature. As the printing speed is changed, it may be necessary to change the temperature of the print head to ensure that the extruded filament is heated sufficiently for deposition. Any suitable print head temperature may be used in some embodiments.

Interface 300 may also include fields to control fill density and fill pattern settings.

Fill density refers to the structure printed within an object and can be specified, for example, by a percentage. Any suitable packing density may be used in some embodiments. 100% packing density refers to a solid packing density with no expected interstices.

Fill pattern refers to the pattern of filling. Any suitable fill pattern may be used in some embodiments. For example, in some embodiments, fill patterns may include honeycombs, triangles, grids, and rectangles. Fill density and fill pattern affect print weight, print object strength, total print time, and exterior properties. Fill patterns can also affect mechanical, optical and/or electrical properties. In some embodiments, the fill density and fill pattern can be set for specific layers or for the entire object.

Additionally, interface 300 may include fields for controlling support settings, including support type and platform binding type.

Some printed objects may have overhangs, so supports may be required to prevent the extruded filament from falling during printing. Support settings can be used to specify where supports are placed. Any suitable support settings may be used in some embodiments.

The platform adhesion setting can be used to improve the adhesion of the printing layer to the bottom plate 150 . The different types of platform bonding settings that can be specified include: Raft settings, which add extra wires in the form of a coarse grid between the bottom layer of the printed object and the base plate; Edge settings, which add extra wires around the first layer of the printed object Lines; and the Skirt setting, which adds a row of wires around objects on the first print layer. Using certain platform binding settings can help reduce the amount of warping in printed objects. Any suitable platform binding settings may be used in some embodiments.

Interface 300 may also include field settings related to object placement and orientation. These settings include the position of the printed object on the bottom plate 150 and the direction of the printed object on the bottom plate 150 . Any suitable object placement and/or orientation may be used in some embodiments.

In some embodiments, the generated digital control code may describe a print path for relative movement of the print head 140 and/or the chassis 150 . A print path can be defined by two sets of coordinates in the X-Y-Z dimension (setpoints), and instructions specifying how to move between the setpoints. The resulting digital control code can specify the speed at which print head 140 and/or chassis 150 move relative to one another (while moving between a pair of consecutive points along the print path), the speed at which the wire will move between a pair of consecutive points. temperature (or print head nozzle temperature), and/or the feed rate of the filament between a pair of consecutive points. The resulting digital control code may specify where the print head 140 should extrude the filament, and/or where it should avoid releasing the filament. All parameters included in the resulting numeric control code are also considered "printing parameters".

One or more of the above printing parameters as well as uncontrollable variables can affect the properties of the printed object, such as layer anomalies, surface roughness, printing resolution, total build time, amount of extruded material used to print the object , and the mechanical, optical and/or electrical properties of the printed object. Mechanical properties may include maximum tensile strength (R _m ), yield strength (Rp _2% ), elongation at break (A _% ), Young's modulus (E), fatigue (σ _d ), Poisson's ratio , mass, and specific gravity. Optical properties may include absorption, reflection, transmission, and refraction. Electrical properties may include resistivity and conductivity. The disclosed mechanical, optical and electrical properties are examples only and are not intended to be limiting.

AIFC can be used to optimize desired mechanical properties, optical properties, electrical properties, and/or any suitable properties of a printed object. AIFC can also be used to take corrective action when printing objects. Corrective actions may include changing the printing parameters of the next or targeted future layers of the object currently being printed. In some embodiments, AIFC can be used to improve production design.

As shown in FIG. 1 , the additive manufacturing system 100 can include an image generator 170 that can process captured images and/or videos of printed layers of an object. In some embodiments, the image generator 170 may include a three-dimensional topographical image configured to store captured images and/or videos and to generate printed layers from the captured images and/or videos, and/or other suitable Image hardware or software.

Different topographical imaging techniques (including, but not limited to, focus shaping algorithms, light shadow reconstruction algorithms, optical stereo algorithms, and Fourier stack diffraction modulation algorithms) with illumination lights of predefined sizes, quantities, and positions can be used to generate each One or more 3D topography images of the print layer. The resulting topographical image may provide volume information related to the finished layers of the printed object and/or partially printed object, the overall dimensions of each layer of the printed object and/or partially printed object, the printed object and/or the partial The characteristics of each layer of the printed object, and information about anomalies (such as amount, distribution, type of anomaly, etc.) found on one or more layers of the printed object and/or part of the printed object.

An example of a focus shaping algorithm that may be adapted for use in image generator 170 in additive manufacturing system 100 is described by Said Pertuz et al. ("Analysis of Focus Measure Operators for Shape-from-Focus," Pattern Recognition, vol. 45, issue 5, pp. 1415-1432), the entire contents of which are hereby incorporated by reference. The disclosed methods are examples only and are not intended to be limiting.

An example of a light and shadow reconstruction algorithm that can be applied to the image generator 170 in the additive manufacturing system 100 is given by Byungil Kim et al. (“Depth and Shape from Shading using the Photometric Stereo method,” CVGIP: Image Understanding, vol. 54, no. 3 , pp 416-427, 1991), the entire contents of which are hereby incorporated by reference. The disclosed methods are examples only and are not intended to be limiting.

An example of an optical stereo algorithm applicable to image generator 170 in additive manufacturing system 100 is given by Jose R.A. Torreao (“Estimating 3-D Shape from the Optical Flow of Photometric Stereo Images,” Proceedings of the 6th Ibero-American Conference on AI : Progress in Artificial Intelligence (IBERAMIA 1998), Helder Coelho (Ed.), Springer-Verlag, London, UK, UK, 253-261) described in "Estimating 3D shape from optical flow in optical stereo images", the entirety of which can be found at This is incorporated by reference. The disclosed methods are examples only and are not intended to be limiting.

An example of a Fourier stack diffraction modulation algorithm that may be adapted for the image generator 170 in the additive manufacturing system 100 is given by Guoan Zeng et al. ("Wide-field High-resolution Fourier Ptychographic Microscopy," Nature Photonics, vol. 7, pp. 739-745, 2013), the entire contents of which are hereby incorporated by reference. The disclosed methods are examples only and are not intended to be limiting.

In some embodiments, topographical images and/or other suitable images generated by image generator 170 may provide information such as anomaly rates and distributions, anomaly types, deposited wires at various points along the print path, and the like. For example, as shown in FIG. 4, a captured image 400 of a printed layer shows an unexpected gap in the deposited wire. In another set of examples, as shown in Figures 5A, 5B, and 5C, captured images of printed layers showed unexpected thread-like artifacts and other breakages in the printed layer.

In some embodiments, image analyzer 180 may be configured to receive generated topographical images and/or other suitable images from image generator 170 and to visually recognize and identify one or more anomalies on the printed layer . In some embodiments, this can be accomplished by comparing, for example, the actual characteristics of the printed layer obtained from the 2D or 3D topographical image, the printed path map of the detected printed layer, and and/or the layer image; and the characteristics of the printed layer as specified in the generated digital code and/or the layer's production design. In some embodiments, one or more artificial intelligence algorithms may be used to identify anomalies based on differences. These anomalies can include, for example, differences between the actual printed layer and the production design, and/or printed layers related to: perimeter dimensions of layers; dimensions of deposited wire between setpoints; fill density; fill pattern; surface roughness; print path; and/or any other variation of the resulting numeric code. Identification of the anomaly may include classifying the anomaly, as well as identifying its size, shape, X-Y-Z location, and/or any other suitable characteristics. In some embodiments, any suitable artificial intelligence algorithm may be used. For example, in some embodiments, the artificial intelligence algorithm may include one or more of the following alone or in combination: machine learning; hidden Markov model; recurrent neural network; convolutional neural network; Bayesian notation ; general adversarial networks; support vector machines; and/or any other suitable artificial intelligence algorithm.

In some embodiments, algorithms that are not based on artificial intelligence may be used to identify anomalies.

In some embodiments, the image analyzer 180 may be preprogrammed to identify certain anomalies in the received image of the printed layer (e.g., unexpected gaps or curled edges, warping, or uneven patterns in the printed layer). , over-extruded points, corrugations or other foreign artefacts and/or any other ruptures). Based on the preprogrammed anomalies, the image analyzer 180 can process the resulting image of the completed printed layer to determine whether the processed image includes any anomalies similar to the preprogrammed anomalies, and to identify any such anomalies on the printed layer. One or more locations of the exception.

In some embodiments, image analyzer 180 may also be configured to determine and record a correlation between detected anomalies and one or more printing parameters. For example, by using suitable artificial intelligence algorithms, the image analyzer can determine one or more printing parameters that may be related to the detected anomaly. For example, the image analyzer 180 may find the following example dependencies: interruptions that occur when the print head is at certain temperatures but not others; certain print speeds but not others cause a large number of unexpected gaps; Patterns, at certain locations on objects, cause certain types of anomalies.

In some embodiments, in response to detecting a correlation between the detected anomaly and one or more printed patterns, the image analyzer 180 may provide for changing one or more layers of an object, or a pattern to be printed in the future. Information, data and/or instructions on the manner in which one or more objects are printed. For example, in some embodiments, the image analyzer may communicate the found correlations and/or instructions for adaptively adjusting printing parameter settings to the digital control code generator 110, the control module 160, and/or any other device . This information can then be used by the digital control code generator 110 and/or the control module 160 to make adjustments to the printing parameters in the digital control code for any subsequent layers of the object currently being printed. In some embodiments, printing parameters can be adjusted so that the next layer or any future layer compensates for anomalies found in previous layers. For example, if an unexpected gap is detected in a printed layer, the next layer's digital control code may include instructions to fill the gap when depositing wire over the gap. In another example, when an unexpected gap is found in the lower part of the object, the digital control code of the symmetric layer in the upper part of the object may include instructions to compensate for the gap.

In some embodiments, image analyzer 180 may be configured to measure mechanical, optical, and/or electrical properties of the completed printed object.

In some embodiments, image analyzer 180 may be configured to detect correlations between one or more printing parameters and minor anomalies in the printed layer and/or the finished printed object. In a further embodiment, the image analyzer 180 may be configured to detect a correlation between one or more printing parameters and measured mechanical, optical and/or electrical properties of the finished printed object. In response to detecting one or more such correlations, image analysis may provide information, data and/or instructions that alter one or more layers of an object or one or more layers to be printed in the future. The way an object is printed. In some embodiments, image analyzer 180 may provide information, data and/or instructions to, for example, 3D modeling software to improve production design.

In some embodiments, image analyzer 180 may use generated topographical images and/or other generated images for printed layers, and generated numerical control codes for printed layers, to learn uncontrollable variables ( That is, the relationship between uncontrollable variables without human intervention) and the resulting print head motion, as well as learning anomalies in the deposited layer (e.g., unexpected gaps or curled edges in the printed layer, warping or not Uniform patterns, over-extruded dots, deviations from the print path specified in the NC code, unintended threading or other foreign artifacts and/or any other breakage). In response to detecting correlations and anomalies between uncontrollable variables and resulting print head motion, image analyzer 180 may provide changes to one or more layers of an object or one or more objects to be printed in the future to be printed. information, data and/or instructions in the form of

In some embodiments, information about correlations may be used to train one or more AI mechanisms as described herein.

In some embodiments, after printing a layer, the image analyzer 180 can be configured to compare the completed printed layer with one or more previous layers to detect and record anomalies, compare and record anomaly rates and patterns, And provide instructions to the digital control code generator 110 and/or the control module 160 to adjust the printing parameters to optimize the overall design of the object (for example, to obtain desired mechanical, optical and/or electrical properties or to achieve very similar print design for production design) or operations to optimize the print job (for example, to speed up the deposition rate, or minimize the amount of material required). Comparisons between completed layers and previous layers to identify anomalies can also be used to better assign causal relationships to the printing parameters of the additive manufacturing system 100, and to perform the next or any subsequent layers of a partially printed object. Appropriate adjustments and optimization for future print jobs of similar or different objects.

In some embodiments, image analyzer 180 may be configured to analyze the total anomaly rate of the current and/or previous layers of a partially printed object and, based on the AIFC from a similar print job, provide information to digital control code generator 110 and/or Or the control module 160 provides instructions to adjust the printing parameters of the next layer and/or any future layers of the partially printed object to obtain desired mechanical, optical and/or electrical properties.

In some embodiments, after printing a layer, non-artificial intelligence based algorithms may be used to identify anomalies related to the specific calibration of the specific additive manufacturing printer being used. Appropriate adjustments can be made to the digital control codes for subsequent layers and future print jobs to account for the calibration of a particular additive manufacturing printer.

In some embodiments, if anomalies in the print layer or layers exceed some predetermined tolerance, the print job of the printed object may be stopped before completion. The data collected for failed print jobs may provide information to digital control code generator 110, control module 160, and/or any computer system that collects training data related to the printing process performed by additive manufacturing printer 115, Information and/or Instructions.

In some embodiments, image analyzer 180 may also be configured to communicate anomaly data of printed objects (e.g., distribution, pattern, and velocity of anomalies), as well as automatically recommended printing adjustments to 3D modeling software to eliminate Or modify the structure in the design corresponding to the exception.

With further reference to Figures 1-3, Figure 6 shows an example of a high-level, additive manufacturing printing operation using an AIFC, according to some embodiments of the disclosed subject matter. In some embodiments, the additive manufacturing process 600 may use the additive manufacturing system 100 .

At 610 , the digital control code generator 110 is provided with a production design specifying what the printed object should look like, as well as desired mechanical, optical, and/or electrical properties of the printed object. In some embodiments, some initial printing parameters are input by the operator. In some embodiments, the production design is provided to the digital control code generator 110 and the image analyzer 180 using AIFC to determine the desired mechanical, optical and/or electrical properties of the production design.

In some embodiments, an operator may input a set of rules for the image analyzer 180 to resolve conflicting goals during the additive manufacturing printing process. For example, users can specify that achieving the best mechanical properties for printed objects should take precedence over: printing speed; integrity of design for production; and reducing the amount of filament used. The operator can also specify the most important mechanical, optical, and/or electrical properties of the printed object so that the image analyzer 180 can provide instructions for adjusting printing parameters that optimize those mechanical, optical, and/or electrical properties .

At 620, the numerical control code generator 110 may generate a numerical control code for printing the object layer based on one or more of: input parameters entered by the operator, printing characteristics of the additive manufacturing printer 115; Specifications of the production design (including mechanical, optical and/or electrical properties); AIFC from one or more previously printed layers of a partially printed object and/or AIFC from other printed objects. The generated digital control code may include a set of set points (eg, a plurality of X-Y-Z coordinates) for print head 140 and/or chassis 150 to travel through. Figure 7A shows an example simulation of a set of set points for a print layer that might be included in a digital control code. The generated digital control code may also include instructions defining how the print head and/or the plate should travel past various setpoints. For example, an example simulation of how the print path traveled might look based on the instructions included is shown in Figure 7B.

In some embodiments, the generated digital control code may also specify certain printing parameters, including but not limited to, the speed of the print head 140 and/or the chassis 150 between set points, the speed of the print head 140 between set points, The temperature of the nozzle and/or the base plate 150, the amount of wire deposited between the set points, the fill density between the set points, and the fill pattern between the set points.

At 630 , the printhead 140 may deposit wires for producing the designed layers according to the instructions provided by the digital control code generator 110 and/or the control module 160 .

At 640, the light source 130 may illuminate the print layer based on the specified topographical imaging technique and/or (as described above) other imaging techniques.

At 650, the image sensor 120 may capture an image of the illuminated print layer.

At 660 , image generator 170 may generate one or more topographical images of the print layer and/or any other suitable images based on the images captured by image sensor 120 . In some embodiments, the generated image of the printed layer may include a series of captured images tiled or concatenated together.

In a further embodiment, the actual print path of the print layer may be determined from one or more topographical images generated by image generator 170 and/or other suitable images.

At 670, the image analyzer 180 may use the generated topographical image and/or other generated images of the printed layer, and the generated digital control code of the printed layer to determine and record anomalies in the extruded layer (e.g., unintended gaps or curled edges, warped or uneven patterns, over-extruded dots, deviations from the print path specified in the NC code, unintended corkscrews or other extraneous artifacts and/or or any other break in the print layer).

In some embodiments, the image analyzer 180 may extract and map the print path of the print layer from the setpoints and instructions included in the generated digital control code. The image analyzer may convert the drawn print path to pixels and overlay the pixels on the print path obtained from the generated image of the print layer and determine differences between the pixels and the print path. In some embodiments, the image analyzer 180 may convert the print path obtained from the generated image of the print layer into print points in a coordinate system, and compare these print points along the lines extracted from the generated digital control code. Take the printed points of the drawn path for comparison.

If the print path of the print layer obtained from the generated image is the same as the print path extracted from the generated NC code, the difference between them will be zero or close to zero. A number greater than zero describes the amount of errors detected between the actual print path and the print path specified in the generated NCC. The comparison of print paths can also point out where errors occur along the print path.

At 680, the image analyzer 180 can analyze the number and type of anomalies detected by the image analyzer from the print layer and/or the previous layer (both the actual path and the print path in the generated digital control code deviation between). Based on the AIFC from other print jobs, the image analyzer 180 can determine whether any adjustments should be made to the printing parameters of the next or subsequent layer of the partially printed object to achieve the desired mechanical, optical and/or electrical properties. For example, if image analyzer 180 determines that the mechanical properties of the completed printed object will be weaker than expected for the current and/or previous layers of the partially printed object based on detected anomalies, image analyzer 180 may instruct digital control Code generator 110 and/or control module 160 adjust certain printing parameters (eg, increasing fill density and/or changing fill pattern) on the next or any subsequent layer so that desired mechanical properties can be achieved.

In some embodiments, operations 610-680 are repeated for each layer or any number of layers deposited for the printed object. The image analyzer 180 can use the data obtained at each layer and the AIFC from other print jobs to modify the printing parameters of the next and/or subsequent layers to achieve desired mechanical, optical, and/or electrical properties of the printed object and/or desired design. In a further embodiment, mechanical, optical and/or electrical properties of the finished printed object may be measured.

The division of time for performing particular portions of process 600 may vary, and no division or different divisions are within the scope of the subject matter disclosed herein. Note that in some embodiments, the blocks of process 600 may be performed at any suitable time. It should be understood that in some embodiments, at least some portions of the process 600 described herein may be performed in any order or sequence, not limited to the order and sequence shown and described in FIG. 6 . Furthermore, in some embodiments, some portions of process 600 described herein may be performed substantially simultaneously, as appropriate or in parallel. Additionally or alternatively, some portions of process 600 may be omitted in some embodiments.

The process 600 can be implemented by any suitable hardware and/or software. For example, in some embodiments, the process 600 may be implemented in the image analyzer 180 or the digital control code generator 110 .

In some embodiments, for similar or dissimilar objects, image analyzer 180 may learn anomalous patterns for each layer of the printed object, such that image analyzer 180 can identify (eg, as described in connection with FIG. Printing parameters are adaptively adjusted at the layer level during the printing process to achieve desired mechanical, optical and/or electrical properties.

Certain printing parameters affect the mechanical, optical and/or electrical properties of the printed object. For example, packing density and packing pattern can affect mechanical properties such as maximum tensile strength (R _m ), yield strength (Rp _2% ), elongation at break (A _% ), Young's modulus (E), fatigue (σ _d ), Poisson's ratio, mass and specific gravity.

In order to understand how anomalous patterns and certain printing parameters actually affect the mechanical, optical and/or electrical properties of the printed object, it is possible to print the object multiple times while changing the mechanical, optical and/or electrical properties of the printed object print parameters for . Anomalous patterns for each printed object may be determined and recorded at the layer level (eg, at 670 ), such as described in connection with FIG. 6 . Additionally, mechanical, optical and/or electrical properties of each printed object can be measured and recorded.

FIG. 8 shows an example 800 of a training process for learning outlier patterns and outlier rates based on different infill densities and infill patterns, and how those anomaly patterns and outlier rates affect the mechanical properties of printed objects, according to some embodiments.

At 810, one or more printing parameters that affect mechanical properties of the object can be identified. For example, it can be recognized when the infill density and infill pattern affect the mechanical properties of the object. In some embodiments, artificial intelligence algorithms can be used to identify other printing parameters that affect the mechanical properties of the object.

At 820, a predetermined number ("group") of target objects may be printed, and one or more printing parameters that have been identified to affect mechanical properties of the target objects may remain the same for the entire group. For example, you can print each object in a group, specifying the same hatch pattern and fill density print parameters.

At 830, as described in connection with 670 of FIG. 6, anomalies may be detected and logged at the layer level for each printed object in the group. For example, the difference between the actual print path and the print path extracted from the generated control code can be determined for each layer. After printing each target object in the set, the mechanical properties of the target object can be measured and recorded.

After 830, process 800 may loop back to 820 and another predetermined number of target objects may be printed with a different fill density and/or fill pattern.

820 and 830 may be repeated as many times as necessary to train image analyzer 180 to learn how anomalous patterns and different identified printing parameters (eg, infill density and infill pattern) affect the mechanical properties of the object. The identified parameters (eg, infill density and/or infill pattern) may be changed each time a predetermined number of target objects ("set") are printed. The table below reflects an example set of target objects and their specified infill density and infill pattern print parameters:

In some embodiments, the fill density and fill pattern printing parameters are maintained constant for each layer of the printed object. In other embodiments, the fill density and/or fill pattern is varied for different layers of the object, depending on the position of the layer or randomly.

Once the image analyzer 180 has learned how different outlier rates and patterns and identified print parameters (e.g., different infill densities and infill patterns) affect the mechanical properties of the object, during a print job (e.g., at the layer level ) image analyzer can adaptively adjust the values of the identified printing parameters to achieve the desired mechanical properties. For example, the image analyzer 180 may detect that the printed layer of a partially printed object has certain anomalies and patterns that, if the fill density and fill pattern are not adjusted, may result in a lower-than-average printed object once completed. mechanical properties. The image analyzer 180 can then adjust the fill rate and fill pattern printing parameters of the next layer and/or any subsequent layers to achieve the desired mechanical properties while also attempting to reduce the occurrence of anomalies.

Similar procedures can be performed to learn how anomalous patterns affect the optical and/or electrical properties of an object. For example, printing parameters that affect the optical and/or electrical properties of the object can be identified. Multiple sets of target objects can be printed by controlling the printing parameters for identifying multiple sets of target objects as described above. Once the image analyzer 180 has learned how the different anomaly rates and patterns and the identified printing parameters affect the electrical and/or optical properties of the object, the image analyzer can adaptively adjust all the anomalies at the layer level during the printing job. Values of the identified printing parameters to achieve desired electrical and/or optical properties.

Similar procedures can be performed for learning how uncontrollable variables (ie, uncontrollable variables without human intervention) affect mechanical, optical and/or electrical properties of objects. For example, uncontrollable variables affecting mechanical, optical and/or electrical properties of an object may be identified. By controlling the identified uncontrollable variables of multiple sets of target objects as described above, multiple sets of target objects can be printed. Once the image analyzer 180 has learned how the different anomaly rates and patterns and identified printing uncontrollable variables affect the mechanical, electrical, and/or optical properties of the object, the image analyzer can automatically, at the layer level, during a printing job. The values of printing parameters are adaptively adjusted to compensate for uncontrollable variables and achieve desired electrical and/or optical properties.

The division of time for performing particular portions of process 800 may vary, and no division or different divisions are within the scope of the subject matter disclosed herein. Note that in some embodiments, the blocks of process 800 may be performed at any suitable time. It should be understood that in some embodiments, at least some portions of the process 800 described herein may be performed in any order or sequence, not limited to the order and sequence shown and described in FIG. 8 . Furthermore, in some embodiments, some portions of process 800 described herein may be performed substantially simultaneously, as appropriate or in parallel. Additionally or alternatively, some portions of process 800 may be omitted in some embodiments.

The process 800 can be implemented by any suitable hardware and/or software. For example, in some embodiments, the process 800 may be implemented in the image analyzer 180 or the digital control code generator 110 .

_{i -1} 表示）；造成列印頭的先前位置的控制碼列印參數（以θ_{i -1} 表示）；以及列印頭的目前測量位置（以

_i 表示）。輸出變量可以是造成列印頭目前位置的數字控制碼參數（以θ 表示）。輸入變量及輸出變量可以一起作為人工智慧算法的單一訓練樣本。單一列印層可以產生數百個這樣的訓練樣本。這些訓練樣本，以及先前層中的異常知識、生產設計的期望規格、積層製造印表機的列印特徵及/或環境條件可用於計算最佳列印參數以產生期望的列印頭運動。在一些實施例中，訓練樣本、以及先前層中的異常知識、生產設計的期望規格、積層製造印表機的特徵及/或環境條件可用於計算最佳列印參數、以及在後續層中用於列印路徑的X-Y-Z設定點的最佳位置及指令。In some embodiments, image analyzer 180 may use generated topographical images of printed layers and/or other generated images (as described in Figure 6 660), and generated digital control codes of printed layers, to study the relationship between printing parameters and the resulting print head motion, and to study anomalies in extruded layers (e.g., unexpected gaps or curled edges in the printed layer, warped or uneven patterns, excessive extruded dots, curled edges, deviations from the print path specified in the NC code, unintended threading or other foreign artifacts, and/or any other breakage). Image analyzer 180 can also invert the learned relationship to calculate optimal digital control code input parameters that will result in desired print head motion and minimize anomalies in the extruded layer. More specifically, the input variables for the AI algorithm may include: the previously measured position of the print head (in

_{i -1} ); the control code printing parameters that caused the previous position of the print head (denoted by θ _{i -1} ); and the current measured position of the print head (denoted by

_i means). The output variable may be a numerical control code parameter (denoted by θ ) that results in the current position of the print head. The input variables and the output variables can be used together as a single training sample for the artificial intelligence algorithm. A single print layer can generate hundreds of such training samples. These training samples, along with knowledge of anomalies in previous layers, desired specifications of production designs, printing characteristics of additive manufacturing printers, and/or environmental conditions can be used to calculate optimal printing parameters to produce desired print head motions. In some embodiments, training samples, as well as knowledge of anomalies in previous layers, desired specifications for production designs, additive manufacturing printer characteristics, and/or environmental conditions, can be used to calculate optimal printing parameters and use them in subsequent layers. Optimal location and command of the XYZ setpoints on the print path.

In some embodiments, the image analyzer 180 can also be used to learn the relationship between printing parameters and the overall characteristics of the layer. For example, image analyzer 180 may be employed to learn the total number of anomalies in extruded layers, the proximity of print head and/or platen motion to the print path instructions in the generated digital control code, and specified fill densities. In some embodiments, image analyzer 180 may invert the learned relationship to calculate printing parameters that will result in a layer that most closely matches the specifications of the production design, including desired mechanical, optical, and/or electrical properties.

In some embodiments, image analyzer 180 may use topographical images and/or other generated images of printed layers, as well as digital control codes generated by printed layers, to learn uncontrollable variables (i.e., no uncontrollable variables of human intervention) and the resulting print head motion, and learning anomalies in the deposited layer (e.g., unexpected gaps or curled edges in the printed layer, warped or uneven patterns, over-extruded points, deviations from the print path specified in the NC code, unintended threading or other foreign artifacts, and/or any other breakage). If the image analyzer 180 finds that uncontrollable variables adversely affect the resulting print head motion and/or anomalies in the deposited layer (beyond a threshold tolerance), the image analyzer 180 may send an alert to the control module 160 . Once an alert is received, the control module 160 may display an alert on a display of the additive manufacturing system 100 and/or alert an operator via email, text, or any other suitable electronic mechanism. In some embodiments, imagery analyzer 180 may be configured to alert the operator directly via email, text, or any other suitable electronic mechanism. For example, in some embodiments, if image analyzer 180 determines that ambient humidity, temperature, and/or light are negatively affecting the resulting print head motion or that the number of anomalies in a layer exceeds a predetermined tolerance, image analyzer 180 may send a message to The control module 160 and/or the operator sends an alert. In some embodiments, if image analyzer 180 determines that wear of additive manufacturing printer 115 and/or the amount of bus material (eg, a small amount of wire) available to print head 140 negatively affects the resulting print head movement or layer The image analyzer 180 may send an alert to the control module 160 and/or an operator to replace the AM printer and/or refill the wire if the number of anomalies in the image exceeds a predetermined tolerance. In some embodiments, if image analyzer 180 determines that voltage variations are negatively affecting the resulting print head motion or that the number of anomalies in a layer exceeds a predetermined tolerance, image analyzer 180 may report to control module 160 and/or operate or send an alert to check the voltage source.

In some embodiments, any suitable computer-readable medium may be used to store instructions for performing the functions and/or processes described herein. For example, in some embodiments computer readable media may be transitory or non-transitory. For example, non-transitory computer-readable media may include media such as non-transitory magnetic media (such as hard disks, floppy disks, etc.), non-transitory optical media (such as compact discs, DVDs, Blu-ray discs, etc.), non-transitory semiconductor media (such as flash memory, electrically programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.), will not be short or lack any persistence during transmission any suitable medium of representation, and/or any suitable tangible medium. As another example, transitory computer-readable media may include signals on a network, wires, conductors, optical fibers, circuits, and any suitable medium that is transitory or lacks any appearance of permanence during transmission, and/or any Signals in suitable intangible media.

The provision of examples described herein (and clauses expressing "such as", "for example", "including", etc.) should not be construed as limiting the claimed subject matter to the particular examples; rather, these examples are for illustration only Some of the many possible aspects. It should also be noted that, as used herein, the term mechanism can encompass hardware, software, firmware, or any suitable combination thereof.

The additive manufacturing system and method have been described in detail with specific reference to these illustrated embodiments. However, it is obvious that various modifications and changes can be made within the spirit and scope of the present disclosure as described in the foregoing specification, and these modifications and changes are considered to be equivalents of the present disclosure and a part of the present disclosure. The scope of the present invention is limited only by the claims of the following claims.

110‧‧‧Digital control code generator 120‧‧‧image sensor 130‧‧‧Light source 140‧‧‧print head 145‧‧‧Wire supply system 150‧‧‧bottom plate 160‧‧‧control module 170‧‧‧Image generator 180‧‧‧Image Analyzer 210‧‧‧Light-emitting diode (LED) ring 230‧‧‧LED bracket 300‧‧‧interface 400‧‧‧capturing images 600, 800‧‧‧Process Steps 610, 620, 630, 640, 650, 660, 670, 680, 810, 820, 830‧‧‧

Figure 1 is an example of an additive manufacturing system according to some embodiments. Figure 2 is an example of a camera and light source that may be used with an additive manufacturing printer according to some embodiments. FIG. 3 is an example of an interface for an operator to input printing parameters into a digital control code generator according to some embodiments. FIG. 4 is an example of an image of a printed layer that may be captured according to some embodiments, showing unexpected gaps in the deposited wire. Figures 5A, 5B, and 5C are examples of various print layer images that may be captured according to some embodiments, showing unexpected thread-like artifacts and other breakages in the print layer. Figure 6 is an example of a (higher order) additive manufacturing printing operation according to some embodiments. Figure 7A is an example of a simulation of a set of set points for a print layer that may be included in a digital control code according to some embodiments. Figure 7B is an example of a simulation of how a traveled print path might look according to some implementations. Figure 8 is an example of a training process for learning outlier patterns and outlier rates based on different infill densities and infill patterns, and how these outlier patterns and outlier rates affect the mechanical properties of printed objects, according to some embodiments.

110‧‧‧Digital control code generator

120‧‧‧image sensor

130‧‧‧Light source

140‧‧‧print head

145‧‧‧Wire supply system

150‧‧‧bottom plate

160‧‧‧control module

170‧‧‧Image generator

180‧‧‧Image Analyzer

Claims (21)

An additive manufacturing system comprising: a printing head configured to print an object in a layer-by-layer manner; an illumination source for providing illumination to a surface of a printed layer of the object; configured to capture an image of the printing layer; and at least one hardware processor configured to: receive a captured image; obtain a target mechanical property, a target optical property, or a target electrical property of the object at least one of; generating a three-dimensional topographical image of the printed layer; identifying the printed layer from the generated topographical image using a first artificial intelligence algorithm trained to detect anomalies in the printed layer an anomaly in; using a second artificial intelligence algorithm, determining a correlation between the identified anomaly, one or more human uncontrollable variables, and one or more printing parameters; determining that the object is no longer reachable based on the correlation at least one of the target mechanical property, the target optical property, or the target electrical property; adjusting a value of the one or more printing parameters to achieve the target mechanical property, the target optical property, or at least one of the target electrical properties; prompting an operator to adjust a value of the one or more personal uncontrollable variables to achieve at least one of the target mechanical properties, the target optical properties, or the target electrical properties of the object one of; and causing the print head to print a subsequent layer of the object using adjusted values to achieve at least one of the target mechanical property, the target optical property, or the target electrical property of the object. The additive manufacturing system described in item 1 of the scope of patent application, wherein the three-dimensional shape image of the printing layer is generated using one of the following algorithms: a shape-from-focus algorithm algorithm), a shape-from-shading focus algorithm, a photometric stereo algorithm, and a Fourier ptychography modulation algorithm. The additive manufacturing system described in claim 1, wherein at least one of the first artificial intelligence algorithm and the second artificial intelligence algorithm includes machine learning, hidden Markov models, recursive class In recurrent neural networks, convolutional neural networks, Bayesian symbolic methods, support vector machines, and general adversarial networks at least one of the . The additive manufacturing system of claim 1, wherein identifying the anomaly is performed by comparing the generated three-dimensional topography image with at least one of: the generated number for the printed layer The control code, one or more front layers of the printed object, and a production design of the printed object. The additive manufacturing system described in claim 1, wherein the at least one hardware processor is further configured to measure at least one of the mechanical properties, the optical properties, and the electrical properties after the object is printed determining at least one of an anomalous rate and an anomalous pattern of the object; and determining how the anomalous rate and the anomalous pattern and at least one of the printing parameters affect the mechanical property, the optical properties and at least one of the electrical properties. In the additive manufacturing system described in claim 1 of the patent application, an artificial intelligence algorithm is used to identify a printing parameter. The additive manufacturing system described in claim 1, wherein determining how the abnormality rate and the abnormal pattern and at least one of the printing parameters affect at least one of the mechanical properties, the optical properties, and the electrical properties One is to use an artificial intelligence algorithm. A method for additive manufacturing comprising: receiving a captured image produced by an image sensor configured to capture an object printed in a layer-by-layer manner obtaining an image of a printed layer of the object; obtaining at least one of a target mechanical property, a target optical property, or a target electrical property of the object; generating a three-dimensional topographical image of the printed layer using a hardware processor ; using a first artificial intelligence algorithm trained to detect anomalies in the printing layer, identifying an anomaly in the printing layer from the generated topography image; using a second artificial intelligence algorithm, determining the identified anomaly, a correlation between one or more human uncontrollable variables and one or more printing parameters; determining based on the correlation that the target mechanical property, the target optical property, or the target electrical property of the object can no longer be achieved adjusting a value of the one or more printing parameters to achieve at least one of the target mechanical property, the target optical property, or the target electrical property of the object; prompting an operator to adjust a value of the one or more uncontrollable variables to achieve at least one of the target mechanical property, the target optical property, or the target electrical property of the object; and causing the print head to use the adjusted value series A subsequent layer of the object is printed to achieve at least one of the target mechanical property, the target optical property, or the target electrical property of the object. The method described in claim 8 of the patent application, wherein the 3D topography image of the printing layer is generated using one of the following algorithms: a shape-from-focus algorithm, a light and shadow reconstruction algorithm (shape-from-shading focus algorithm), an optical stereo algorithm (photometric stereo algorithm), and a Fourier ptychography modulation algorithm (Fourier ptychography modulation algorithm). The method described in claim 8 of the patent application, wherein at least one of the first artificial intelligence algorithm and the second artificial intelligence algorithm includes machine learning, hidden Markov models, and recurrent neural networks At least one of recurrent neural networks, convolutional neural networks, Bayesian symbolic methods, support vector machines, and general adversarial network one of them. The method of claim 8, wherein identifying the anomaly is performed by comparing the generated 3D topography image with at least one of: a generated digital control code for the print layer , one or more front layers of the printed object, and a production design of the printed object. The method described in item 8 of the scope of the patent application further includes: after the object is printed, measuring at least one of the mechanical property, the optical property and the electrical property; determining an abnormality rate and an abnormality rate of the object at least one of the abnormal pattern; and determining how the abnormal rate and at least one of the abnormal pattern and the printing parameter affect at least one of the mechanical property, the optical property, and the electrical property of the object. The method described in claim 8, wherein identifying a printing parameter uses an artificial intelligence algorithm. The method of claim 8, wherein determining how at least one of the abnormality rate and the abnormal pattern and the printing parameters affect at least one of the mechanical property, the optical property and the electrical property is Use an artificial intelligence algorithm. A non-transitory computer-readable medium containing computer-executable instructions that, when executed by a processor, cause the processor to perform a method for additive manufacturing, the method comprising: receiving a captured image generated by an image sensor configured to capture an image of a printed layer of an object printed in a layer-by-layer manner; obtaining the at least one of a target mechanical property, a target optical property, or a target electrical property of an object; generating a three-dimensional topographical image of the printed layer; using a first an artificial intelligence algorithm identifying an anomaly in the print layer from the generated topography image; using a second artificial intelligence algorithm to determine the identified anomaly, one or more human uncontrollable variables and one or more column a correlation between printing parameters; determining based on the correlation that at least one of the target mechanical property, the target optical property, or the target electrical property of the object is no longer achievable; adjusting the one or more printing parameters a value to achieve at least one of the target mechanical property, the target optical property, or the target electrical property of the object; prompting an operator to adjust a value of the one or more human uncontrollable variables to achieve the object at least one of the target mechanical property, the target optical property, or the target electrical property; and causing the print head to print a subsequent layer of the object using adjusted values to achieve the target mechanical property of the object property, the target optical property, or at least one of the target electrical property. The non-transitory computer-readable medium described in claim 15, wherein the three-dimensional shape image of the printing layer is generated using one of the following algorithms: a shape-from-focus algorithm algorithm), a light and shadow reconstruction algorithm (shape-from-shading focus algorithm), a photometric stereo algorithm, and a Fourier ptychography modulation algorithm. The non-transitory computer-readable medium described in claim 15, wherein at least one of the first artificial intelligence algorithm and the second artificial intelligence algorithm includes machine learning, hidden Markov models (hidden Markov models ), recurrent neural networks, convolutional neural networks, Bayesian symbolic methods, support vector machines, and general adversarial networks ( general adversarial network) at least one of them. The non-transitory computer readable medium of claim 15, wherein identifying the anomaly is performed by comparing the generated 3D topography image with at least one of: for the printed layer The generated digital control code, one or more front layers of the printed object, and a production design of the printed object. The non-transitory computer-readable medium described in claim 15, wherein the method further includes: measuring at least one of the mechanical property, the optical property, and the electrical property after the object is printed ; determining at least one of an anomalous rate and an anomalous pattern of the object; and determining how the anomalous rate and the anomalous pattern and at least one of the printing parameters affect the mechanical property, the optical property, and at least one of the electrical properties. The non-transitory computer-readable medium as described in claim 15, wherein identifying a printing parameter uses an artificial intelligence algorithm. The non-transitory computer readable medium of claim 15, wherein determining how at least one of the anomaly rate and the anomalous pattern and the printing parameters affect the mechanical property, the optical property, and the electrical property At least one of the properties is using an artificial intelligence algorithm.

Images